ggplot2 layer for diagonally split heatmap cells

A direct-to-ggplot interface for the drawing biheat() performs. Expects a long data frame with one row per cell and upper / lower aesthetics holding the two values. Each cell is drawn as two triangles: upper above the diagonal, lower below it.

Usage

geom_biheat(
  mapping = NULL,
  data = NULL,
  stat = "identity",
  position = "identity",
  ...,
  split = c("down", "up"),
  palettes = NULL,
  lower_palette = NULL,
  upper_range = NULL,
  lower_range = NULL,
  cell_inset = 0.04,
  na_colour = "grey90",
  show.legend = NA,
  inherit.aes = TRUE
)

Arguments

mapping: Aesthetic mapping. Required aesthetics are x, y, upper, lower; width, height, colour, linewidth, linetype and alpha are optional.
data: Long data frame, one row per cell.
stat, position: Standard ggplot2 layer arguments.
...: Other arguments passed on to ggplot2::layer(), including fixed aesthetics such as colour or linewidth.
split: "down" (top-left to bottom-right diagonal) or "up".
palettes: A palettes() object, or a character vector of upper colours paired with lower_palette. Defaults to white-to-red over white-to-blue.
lower_palette: Optional companion to a bare palettes vector.
upper_range, lower_range: Numeric length-2 value ranges the palettes are stretched across. Default to the range of the data in each triangle. Pass the same value to both for a shared scale.
cell_inset: Numeric in [0, 0.5). Shrinks each cell on every side to leave a gap between neighbours.
na_colour: Fill for NA values.
show.legend, inherit.aes: Standard ggplot2 layer arguments.

Value

A ggplot2 layer.

Details

The layer carries its own pair of gradient palettes and maps values to colours itself, so the two triangles keep independent colour ranges inside a single ggplot. Neither side consumes the fill scale, which leaves it free for other layers. Add a key with bi_legend(), or use biheat() when you want the legend built for you.

Examples

set.seed(1)
cells <- expand.grid(site = letters[1:5], species = LETTERS[1:4])
cells$observed  <- runif(20)
cells$predicted <- runif(20)

library(ggplot2)
ggplot(cells, aes(site, species, upper = observed, lower = predicted)) +
  geom_biheat() +
  theme_biheat()